We post-trained Ultra for popular agent harnesses like @openclaw
, @NousResearch Hermes Agent, and @Langchain
. The result is an open frontier model developers can customize for specialized agents across domains. Read more: https://
nvda.ws/4adkn6J
LLMS
-
NVIDIA post-trains Ultra for popular agent harnesses
By
–
-
Ultra excels at large codebases, long reasoning chains, and multi-source synthesis
By
–
Beyond benchmark performance, Ultra can work through large codebases, reason across long chains of tool calls, and synthesize information gathered from hundreds of sources. pic.twitter.com/itDu34WVHk
— NVIDIA AI (@NVIDIAAI) 4 juin 2026Beyond benchmark performance, Ultra can work through large codebases, reason across long chains of tool calls, and synthesize information gathered from hundreds of sources.
-
Nvidia AI Ultra model excels at complex tasks with hybrid architecture
By
–
Ultra excels at complex tasks like coding and deep research. Long-running agents spend their time planning, using tools, recovering from failures, and deciding what to do next. The model’s hybrid Mamba-Transformer MoE architecture enables more reasoning cycles within the same
-
Nvidia Nemotron 3 Ultra: 550B MoE open frontier model
By
–
Today we're shipping Nemotron 3 Ultra.
— NVIDIA AI (@NVIDIAAI) 4 juin 2026
A 550B MoE frontier-intelligence open model built for long-running agents.
It delivers 5x faster inference and lowers the cost of complex agentic tasks by up to 30% versus other open frontier models. pic.twitter.com/FEXqvfzQFOToday we're shipping Nemotron 3 Ultra. A 550B MoE frontier-intelligence open model built for long-running agents. It delivers 5x faster inference and lowers the cost of complex agentic tasks by up to 30% versus other open frontier models.
-
OpenAI and Anthropic documentation lags products with obsolete advice
By
–
In general, OpenAI and Anthropic documentation lags their own products by months, and both companies have tons of obsolete and contradictory advice scattered around their sites.
-
Reducing AI Hallucinations: Techniques and Justification with Code Examples
By
–
Full send! If you have techniques to reduce hallucinations, I'm all ears. I ask it to justify each conclusion precisely with examples and references within the code. It works well, but I think we can do better.
-
New metric $/Mt measures revenue per million tokens
By
–
I think a new metric should be $/Mt > money you made per million tokens So let's say you make $2,000/mo and spend 10 million tokens to do that That's $200/Mt ($200 made per million tokens) Because you have a lot of people just spending lots of tokens but it's often
-
AI coding tools expand features but remain poorly documented
By
–
The capabilities of Claude Code and Codex have expanded a lot in recent months, they added many ways to approach work (subagents, skills, goal, workflows, plugins, etc). Given the AI labs can use their own AI to help documentation, a surprising amount is effectively undocumented
-
Copilot SDK can now be used as a harness
By
–
That works for quite a while; new is that you can even use Copilot SDK as harness.
